Thalidomide is a bad analogy for AI risk: it predates modern drug regulation entirely
basedjensen · x · 2026-10-12
In an AI-risk thought-experiment debate, Melinda Chu pushes back on mathematicians using thalidomide as a cautionary historical example: the analogy fails without historical context.
Her argument: thalidomide was developed before modern regulatory and IRB standards. Under today's rules, pregnancy toxicity would be caught in animal studies, and pregnancy testing would be required before human treatment — as the REMS program now mandates. Thalidomide is still used for cancer and autoimmune diseases, with newer same-class drugs like lenalidomide approved. She notes Accutane's iPledge program as a similar modern safety framework.
Related event: Debate Erupts Over AI-Designed Drugs No Human Can Explain(4 posts)→
More from AGI Musings
- Matthew Green: assume models keep improving at the same rate as since early this year — matthew_d_green · 2026-10-12
- LLM edge is lossy corpus and machine speed, not superhuman context — zeeg · 2026-10-12
- AI won't rescue US finances: deficit needs 4.4% growth, CBO says AI adds just 0.1 point — nordicinst · 2026-10-12
- The god computer can cure cancer but the nurse doesn't speak English — wen_ragnarok · 2026-10-12
- Gary Basin: frontier labs' $40T/year path is a Universal Basic Management Consultant — max_paperclips · 2026-10-12
- Consumers want outcomes, not software: the misread premise of the personal software era — _AustinCalvert_ · 2026-10-12